Quantitative Models and Construction Methods - **Model Name**: Maximized Factor Exposure Portfolio (MFE) **Construction Idea**: The MFE portfolio is designed to maximize single-factor exposure while controlling for various real-world constraints such as industry exposure, style exposure, stock weight deviation, and turnover rate. This approach ensures the factor's effectiveness under practical constraints [39][40][41] **Construction Process**: The optimization model is formulated as follows: $\begin{array}{ll}max&f^{T}\ w\\ s.t.&s_{l}\leq X(w-w_{b})\leq s_{h}\\ &h_{l}\leq H(w-w_{b})\leq h_{h}\\ &w_{l}\leq w-w_{b}\leq w_{h}\\ &b_{l}\leq B_{b}w\leq b_{h}\\ &\mathbf{0}\leq w\leq l\\ &\mathbf{1}^{T}\ w=1\end{array}$ - **Objective Function**: Maximize single-factor exposure, where $f$ represents factor values, $f^{T}w$ is the weighted exposure of the portfolio to the factor, and $w$ is the stock weight vector to be solved [39][40] - **Constraints**: - **Style Exposure**: $X$ is the matrix of stock exposures to style factors, $w_b$ is the benchmark weight vector, and $s_l$, $s_h$ are the lower and upper bounds for style factor exposure [40] - **Industry Exposure**: $H$ is the matrix of stock exposures to industries, $h_l$, $h_h$ are the lower and upper bounds for industry exposure [40] - **Stock Weight Deviation**: $w_l$, $w_h$ are the lower and upper bounds for stock weight deviation relative to the benchmark [40] - **Component Weight Control**: $B_b$ is a 0-1 vector indicating whether a stock belongs to the benchmark, $b_l$, $b_h$ are the lower and upper bounds for component weight control [40] - **No Short Selling**: Ensures non-negative weights and limits individual stock weights [40] - **Full Investment**: Ensures the portfolio is fully invested with weights summing to 1 [41] **Evaluation**: This model effectively tests factor validity under real-world constraints, ensuring the factor's predictive power in practical portfolio construction [39][40][41] Quantitative Factors and Construction Methods - **Factor Name**: Specificity **Construction Idea**: Measures the uniqueness of stock returns by evaluating the residuals from a Fama-French three-factor regression [16][19][23] **Construction Process**: - Formula: $1 - R^2$ from the Fama-French three-factor regression, where $R^2$ represents the goodness-of-fit of the regression model [16] **Evaluation**: Demonstrates strong performance in multiple sample spaces, indicating its effectiveness in capturing unique stock characteristics [19][23][25] - **Factor Name**: EPTTM Year Percentile **Construction Idea**: Represents the percentile rank of trailing twelve-month earnings-to-price ratio (EPTTM) over the past year [16][19][23] **Construction Process**: - Formula: Percentile rank of $EPTTM = \frac{\text{Net Income (TTM)}}{\text{Market Cap}}$ over the past year [16] **Evaluation**: Performs well in various sample spaces, particularly in growth-oriented indices [19][23][25] - **Factor Name**: Three-Month Reversal **Construction Idea**: Captures short-term price reversal by measuring the return over the past 60 trading days [16][19][23] **Construction Process**: - Formula: $\text{Return}_{60\text{days}} = \frac{\text{Price}_{t} - \text{Price}_{t-60}}{\text{Price}_{t-60}}$ [16] **Evaluation**: Effective in identifying short-term reversal opportunities, especially in volatile indices [19][23][25] Factor Backtesting Results - **Specificity Factor**: - **Sample Space**: CSI 300 - Weekly Excess Return: 1.18% - Monthly Excess Return: 2.02% - Year-to-Date Excess Return: 4.23% - Historical Annualized Return: 0.51% [19] - **Sample Space**: CSI A500 - Weekly Excess Return: 1.43% - Monthly Excess Return: 2.14% - Year-to-Date Excess Return: 2.71% - Historical Annualized Return: 1.72% [25] - **EPTTM Year Percentile Factor**: - **Sample Space**: CSI 300 - Weekly Excess Return: 0.54% - Monthly Excess Return: 2.01% - Year-to-Date Excess Return: 6.74% - Historical Annualized Return: 3.26% [19] - **Sample Space**: CSI 500 - Weekly Excess Return: 1.01% - Monthly Excess Return: 1.54% - Year-to-Date Excess Return: 1.90% - Historical Annualized Return: 5.24% [21] - **Three-Month Reversal Factor**: - **Sample Space**: CSI 300 - Weekly Excess Return: 0.49% - Monthly Excess Return: 1.35% - Year-to-Date Excess Return: 4.31% - Historical Annualized Return: 1.13% [19] - **Sample Space**: CSI 1000 - Weekly Excess Return: 1.10% - Monthly Excess Return: 2.15% - Year-to-Date Excess Return: 2.59% - Historical Annualized Return: -0.67% [23] Index Enhancement Portfolio Backtesting Results - **CSI 300 Enhanced Portfolio**: - Weekly Excess Return: 0.78% - Year-to-Date Excess Return: 9.31% [5][14] - **CSI 500 Enhanced Portfolio**: - Weekly Excess Return: -0.52% - Year-to-Date Excess Return: 9.90% [5][14] - **CSI 1000 Enhanced Portfolio**: - Weekly Excess Return: 0.07% - Year-to-Date Excess Return: 15.69% [5][14] - **CSI A500 Enhanced Portfolio**: - Weekly Excess Return: 0.26% - Year-to-Date Excess Return: 9.96% [5][14] Public Fund Index Enhancement Product Performance - **CSI 300 Public Fund Products**: - Weekly Excess Return: Max 1.28%, Min -0.98%, Median 0.12% - Monthly Excess Return: Max 4.10%, Min -0.99%, Median 0.61% - Quarterly Excess Return: Max 5.71%, Min -0.90%, Median 1.52% - Year-to-Date Excess Return: Max 9.84%, Min -0.77%, Median 2.87% [31] - **CSI 500 Public Fund Products**: - Weekly Excess Return: Max 1.41%, Min -1.31%, Median 0.04% - Monthly Excess Return: Max 2.56%, Min -0.60%, Median 0.60% - Quarterly Excess Return: Max 5.51%, Min -0.10%, Median 2.60% - Year-to-Date Excess Return: Max 9.88%, Min -0.77%, Median 4.19% [34] - **CSI 1000 Public Fund Products**: - Weekly Excess Return: Max 0.82%, Min -0.47%, Median 0.15% - Monthly Excess Return: Max 3.55%, Min -0.67%, Median 1.07% - Quarterly Excess Return: Max 7.14%, Min -0.58%, Median 3.21% - Year-to-Date Excess Return: Max 15.34%, Min 0.49%, Median 6.75% [36] - **CSI A500 Public Fund Products**: - Weekly Excess Return: Max 1.16%, Min -0.57%, Median -0.04% - Monthly Excess Return: Max 1.89%, Min -1.55%, Median 0.68% - Quarterly Excess Return: Max 3.76%, Min -1.67%, Median 2.20% [38]

Quantitative Models and Factor Construction Factor Names and Construction - **Factor Name: EPTTM One-Year Percentile** - **Construction Idea**: Measures the percentile rank of the earnings-to-price ratio (EPTTM) over the past year to capture valuation trends[6][17] - **Construction Process**: - Calculate the earnings-to-price ratio (EPTTM) for each stock - Determine the percentile rank of the current EPTTM relative to its distribution over the past year[17] - **Evaluation**: Demonstrated strong performance in certain indices like CSI 1000 and CSI All Share, indicating its effectiveness in capturing valuation signals[6][33][47] - **Factor Name: Pre-Expected PEG** - **Construction Idea**: Combines price-to-earnings ratio with expected growth rates to evaluate valuation adjusted for growth[17] - **Construction Process**: - Calculate the price-to-earnings ratio (PE) - Divide PE by the expected growth rate of earnings to derive the PEG ratio - Use analyst consensus forecasts for expected growth rates[17] - **Evaluation**: Exhibited strong performance in indices like CSI 800 and CSI 500, suggesting its utility in growth-adjusted valuation analysis[6][29][33] - **Factor Name: Six-Month UMR** - **Construction Idea**: Captures momentum adjusted for risk over a six-month window[17] - **Construction Process**: - Calculate the cumulative return over the past six months - Adjust for risk using a volatility or beta-based measure - Normalize the adjusted return to derive the UMR score[17] - **Evaluation**: Consistently effective across multiple indices, including CSI 500 and CSI 1000, highlighting its robustness in momentum strategies[6][25][33] - **Factor Name: Standardized Unexpected Earnings (SUE)** - **Construction Idea**: Measures the deviation of actual earnings from analyst expectations, standardized by the forecast error[17] - **Construction Process**: - Calculate the difference between actual and expected earnings - Standardize this difference using the standard deviation of forecast errors - Derive the SUE score for each stock[17] - **Evaluation**: Effective in identifying earnings surprises, with strong performance in CSI 500 and CSI All Share indices[6][25][47] Factor Backtesting Results - **EPTTM One-Year Percentile** - CSI 1000: Weekly return of 1.09%, monthly return of 0.83%, annualized return of 5.54%[6][33] - CSI All Share: Weekly return of 1.09%, monthly return of -0.34%, annualized return of -4.16%[6][47] - **Pre-Expected PEG** - CSI 800: Weekly return of 0.66%, monthly return of 2.66%, annualized return of 3.11%[6][29] - CSI 500: Weekly return of 0.27%, monthly return of 1.02%, annualized return of -1.79%[6][25] - **Six-Month UMR** - CSI 500: Weekly return of 0.76%, monthly return of 1.14%, annualized return of -3.98%[6][25] - CSI 1000: Weekly return of 0.32%, monthly return of 0.08%, annualized return of 2.58%[6][33] - **Standardized Unexpected Earnings (SUE)** - CSI 500: Weekly return of 0.55%, monthly return of -0.06%, annualized return of 1.46%[6][25] - CSI All Share: Weekly return of 0.32%, monthly return of -0.46%, annualized return of -4.36%[6][47] MFE Portfolio Construction - **Model Description**: Maximized Factor Exposure (MFE) portfolios are constructed to maximize exposure to a single factor while controlling for constraints such as industry and style exposures, stock weight limits, and turnover[61][62] - **Optimization Formula**: $$ \begin{array}{ll} \max & f^{T}w \\ \text{s.t.} & s_{l} \leq X(w-w_{b}) \leq s_{h} \\ & h_{l} \leq H(w-w_{b}) \leq h_{h} \\ & w_{l} \leq w-w_{b} \leq w_{h} \\ & b_{l} \leq B_{b}w \leq b_{h} \\ & 0 \leq w \leq l \\ & 1^{T}w = 1 \\ & \Sigma|w-w_{0}| \leq to_{h} \end{array} $$ - **Explanation**: The objective function maximizes factor exposure, subject to constraints on style, industry, stock weights, and turnover[61][62] - **Evaluation**: Effective in isolating factor performance under realistic portfolio constraints, providing a robust framework for factor validation[61][65]